Urban Mobility

SUMO (Simulation of Urban MObility) An open-source traffic simulation Problems and Solutions
Daniel Krajzewicz
Folie 1
SUMO (FU Bozen) > Daniel Krajzewicz > 07_11_14_DLR_dkrajzew_SUMO_FUBozen.ppt > 14.11.2005
Contents
Reasons for an open-source traffic simulation Needed models, data, and its processing SUMO Overview
Folie 2
SUMO (FU Bozen) > Daniel Krajzewicz > 07_11_14_DLR_dkrajzew_SUMO_FUBozen.ppt> 09.11.2005
Open Source Road Traffic Simulation Applications for a road traffic simulation
reproducible, computerized evaluation of

real-world network performance new concepts for
modern traffic signal control traffic surveillance traffic forecasting traffic management dynamic routing methods car2car/car2infrastructure communication
traffic models
instructive visualization planning
Folie 3
Open Source Road Traffic Simulation Needed Components
road network
vehicles / flow
signalling
Folie 4
Open Source Road Traffic Simulation Why an open source road traffic simulation?
Common procedure
An academic organization wants to evaluate an aspect of traffic Builds an own traffic simulation; needs
A network representation including speed limits, right-of-way rules, etc. A representation of vehicles, their movement, routes, etc. A representation of signals, variable message signs, etc.
yields in
Many (incomplete) simulations which results can not be compared
Solution: an extendable simulation as a base for own
development
Folie 5
Open Source Road Traffic Simulation Design criteria
Portability
Achieved by using c++ and portable libraries only
Performance
High execution speed No limitations in network size and number of simulated vehicles
Extendibility Open source

Licensed under GPL Hosted at sourceforge (http://sumo.sourceforge.net)
Microscopic
Each vehicle is modeled explicitly
Folie 6
Road Traffic Simulation Road Traffic Flow Dynamics (one of)
Fundamental diagram of traffic (counted through induction loops)
As interpreted by Kerner
As interpreted by Kim and Keller

Folie 7
Road Traffic Flow Simulation Classes

mesoscopic
macroscopic
microscopic
sub-microscopic
Folie 8
Road Traffic Flow Simulation Microscopic Models
Mostly:
discrete in time Car Following Models: vehicles speed depends on the preceding vehicle
Folie 9
Road Traffic Flow Simulation Car Following Model by Krau*
Features:
continuous in space discrete in time accident-free stochastic driver model
Parameters:
acceleration deceleration max. speed drivers imperfection
[*] Microscopic Modelling of Traffic Flow: Investigation of Collision Free Vehicle Dynamics, S. Krau, DLR (Hauptabteilung Mobilitt und Systemtechnik), 1998, ISSN 1434-8454
Folie 10
Road Traffic Flow Simulation Further Models needed
Lane Changing
Navigation (I have to turn left on next junction) Tactical (left lane allows me to move faster)
Interaction with the network

Stops at red traffic lights Decelerate if other vehicles have a higher right to move on the next junction
Extensions
Vehicle classes, such as busses, which use own lanes Bus stops
Folie 11
Microscopic Road Networks Needed vs. given Information
Needed street attributes:

Number of lanes Allowed speed Per-lane restrictions Allowed continuations
Given: a graph with only few information

Number of lanes (often vague) Allowed speed (often vague)
Needed junction attributes:

Right-of-way Traffic lights program
Folie 12
Microscopic Road Networks Size and Complexity
Additional problems:
very large networks complex junctions
Folie 13
Microscopic Road Networks SUMO Solution: Heuristics for automatic computation

Example: computation of lane-to-lane connections:
1. for each edge: compute turnaround edges 2. for each node: sort each nodes edges 3. for each node: compute each nodes type 4. for each node: set edge priorities 5. for each edge: compute edge-to-edge connections 6. for each edge: compute lanes-to-edge connections 7. for each node: compute lane-to-lane connections 8. for each edge: recheck lanes 9. for each edge: append turnarounds
After 2.
After 5.
After 7.
After all
Folie 14
Traffic Demand Needed Information
Each vehicle is modeled explicitly

mandatory attributes:
ID (name) Vehicle type (including Krau-parameter) Complete route through the network Depart time
optional attributes:
stops
<vehicle id="bus100_west_0d" type="BUS" depart="0" color="1,0,1"> <route>-572658025 -572658026 -572658027 -572658024 </route> <stop bus_stop="west1" duration="20"/> <stop bus_stop="west2" duration="20"/> </vehicle>
Folie 15
Traffic Demand Possible Sources

Real (physical) measures are done at induction loops
Accurate but sparse No information about who is driving from which place to which place, only how many vehicles have passed a certain place
Counts at junctions may provide flows and turning ratios

Less accurate than induction loops and sparse, too mainly for some major inner-city junctions Information about flow spread on junctions
Traffic scientists estimate demands on district level

Even less accurate (estimated), but cover a complete area
Folie 16
Traffic Demand Possible Sources

Real (physical) measures are done at induction loops induction-loop SUMO-routes DFROUTER measures Counts at junctions may provide flows and turning ratios turning ratios flows
JTRROUTER
SUMO-routes
Traffic scientists estimate demands on district level

OD-matrix OD2TRIPS trips DUAROUTER SUMO-routes
Folie 17
SUMO - Modules Included Applications

SUMO: command line simulation GUISIM: simulation with a graphical user interface NETCONVERT: network importer NETGEN: network generator OD2TRIPS: converter for O/Dmatrices
JTRROUTER: router based on turning ratios DUAROUTER: router based on a dynamic user assignment DFROUTER: router which uses detector data
Folie 18
SUMO Features
Microscopic: all vehicles are modeled explicitly Time-discrete, space-continuous car-following model by S. Krau 1 Multi-lane traffic, right-of-way rules Around 100,000 cars in real-time (without graphical output) Traffic lights with time schedules, other traffic management devices Complex networks / Import support Dynamic Routing based on Dynamic User Assignment 2 Other routing modules
[1] Microscopic Modelling of Traffic Flow: Investigation of Collision Free Vehicle Dynamics, S. Krau, DLR (Hauptabteilung Mobilitt und Systemtechnik), 1998, ISSN 1434-8454 [2] Christian Gawron. 1998. Simulation-Based Traffic Assignment. Inaugural Dissertation.
Folie 19
Projects at the DLR where SUMO was used 2002-2006
INVENT
Implementation and verification of traffic management strategies for large urban areas
OIS
Verification of benefits arising from usage of new optical sensors
Traffic Tower
Virtual Traffic Management Environment
WJT2005 / Soccer2006
Integration of standard and airborne detectors into a traffic portal with forecast functionality
TrafficOnline
Traffic surveillance via in-vehicle GSM phones
Folie 20
WJT2005 / Soccer2006 DELPHI Description
Used in the city of Cologne during

The popes visit (world youth day 2005) The world soccer cup 2006
Traffic surveillance using

Highway induction loops Inner-city induction loops Airborne traffic recognition system (mounted on a zeppelin)
Traffic visualization
Integration of gained information into a viewer Presentation for the police
Traffic forecast
30 min into the future using a simulation (extended SUMO)
Folie 21
DELPHI System Overview

Webserver: Tomcat Servlet Container Visualization (Images and traffic) Webbrowser: JavaScript Browser asynchronous XMLRequests
Simulation: Areal extrapolation of measured data Forecast generation (30min) Fed from db using python
Datenbank: MySQL with InnoDB tables No built-in logic
Aggregation/Correction: Guessing missing values Data fusion Done using python scripts
ARGOS: Access via db, webserver
Induction loops:
Floating Car Data:
Folie 22
SUMO Example Results Flow Comparison
Folie 23
SUMO Current and next work
Model extensions
Time steps below 1s New routing applications Consolidated support for vehicle classes (busses, emergency, etc.) Multi-modality / person simulation?
Documentation
Tutorials Improving communication with users via a Wiki
Bug tracking
Better specifications Enlarging the number of acceptance tests
Folie 24
SUMO - Participation
Main development:
Institute of Transportation Systems / DLR
Contribution by:
Centre for Applied Informatics Cologne (-2004) Institute of Telematics at the Lbeck University (2006-)
Project status: Download: Contact:
0.9.7 http://sumo.sourceforge.net Daniel.Krajzewicz@dlr.de

Folie 25

Urban Mobility

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Urban Mobility

Uploaded by

Copyright:

Available Formats

SUMO (Simulation of Urban MObility) An open-source traffic simulation Problems and Solutions

reproducible, computerized evaluation of

instructive visualization planning

Open Source Road Traffic Simulation Needed Components

SUMO (FU Bozen) > Daniel Krajzewicz > 07_11_14_DLR_dkrajzew_SUMO_FUBozen.ppt> 09.11.2005

Solution: an extendable simulation as a base for own

Open Source Road Traffic Simulation Design criteria

Extendibility Open source

Road Traffic Simulation Road Traffic Flow Dynamics (one of)

Fundamental diagram of traffic (counted through induction loops)

As interpreted by Kim and Keller

Road Traffic Flow Simulation Classes

SUMO (FU Bozen) > Daniel Krajzewicz > 07_11_14_DLR_dkrajzew_SUMO_FUBozen.ppt> 09.11.2005

Road Traffic Flow Simulation Microscopic Models

Road Traffic Flow Simulation Car Following Model by Krau*

Road Traffic Flow Simulation Further Models needed

Interaction with the network

Microscopic Road Networks Needed vs. given Information

Needed street attributes:

Given: a graph with only few information

Needed junction attributes:

Microscopic Road Networks Size and Complexity

Microscopic Road Networks SUMO Solution: Heuristics for automatic computation

Traffic Demand Needed Information

Each vehicle is modeled explicitly

Traffic Demand Possible Sources

Counts at junctions may provide flows and turning ratios

Traffic scientists estimate demands on district level

Traffic Demand Possible Sources

Traffic scientists estimate demands on district level

SUMO - Modules Included Applications

Projects at the DLR where SUMO was used 2002-2006

WJT2005 / Soccer2006 DELPHI Description

Used in the city of Cologne during

Traffic surveillance using

DELPHI System Overview

Datenbank: MySQL with InnoDB tables No built-in logic

ARGOS: Access via db, webserver

Floating Car Data:

SUMO Example Results Flow Comparison

SUMO Current and next work

Project status: Download: Contact:

0.9.7 http://sumo.sourceforge.net Daniel.Krajzewicz@dlr.de

You might also like